Robust Variational Bayesian Filter for Systems with Skew t Noise
Shuhui Li, Zhihong Deng, Ruxuan He, Feng Pan, Xiaoxue Feng, Ni Pu · 2020
Considering the pulse interference, measurement outliers and artificial modeling errors, the non-Gaussian heavy-tailed (or skew) noise widely exists in the real environment. However, to data, little literature is related to the state estimation of the system where the process and measurement noises (PMNs) are both expressed as the skew t distribution (STD). To this end, given the hierarchical representation of the STD, a new robust Bayesian filter based on the variational Bayesian (VB) inference is presented to approximately estimate the unknown state via the collected measurements. And an example from the target tracking scenario is given to illustrate the validity of the designed Bayesian filter.